| Message ID | 20190329012913.29538-1-hoegeun.kwon@samsung.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | drm/vc4: Fix with pm_runtime synchronization on DSI | expand |
Hoegeun Kwon <hoegeun.kwon@samsung.com> writes: > There is a problem when often dpms goes from off to on. pm idle is not > in sync and the problem occurs. Modify pm_runtime_put from > asynchronous to synchronous. Why would we need the power domain to go to off before we try to come back? Any idea? Also, please specify what "the problem" is. > > Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> > --- > drivers/gpu/drm/vc4/vc4_dsi.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/vc4/vc4_dsi.c b/drivers/gpu/drm/vc4/vc4_dsi.c > index 11702e1d9011..b0264c7b8bed 100644 > --- a/drivers/gpu/drm/vc4/vc4_dsi.c > +++ b/drivers/gpu/drm/vc4/vc4_dsi.c > @@ -822,7 +822,7 @@ static void vc4_dsi_encoder_disable(struct drm_encoder *encoder) > clk_disable_unprepare(dsi->escape_clock); > clk_disable_unprepare(dsi->pixel_clock); > > - pm_runtime_put(dev); > + pm_runtime_put_sync(dev); > } > > /* Extends the mode's blank intervals to handle BCM2835's integer-only > -- > 2.17.1
On 4/2/19 2:48 AM, Eric Anholt wrote: > Hoegeun Kwon <hoegeun.kwon@samsung.com> writes: > >> There is a problem when often dpms goes from off to on. pm idle is not >> in sync and the problem occurs. Modify pm_runtime_put from >> asynchronous to synchronous. > Why would we need the power domain to go to off before we try to come > back? Any idea? Also, please specify what "the problem" is. Hi Eric, First thank you for your review. There is a problem failed to runtime PM enable on DSI when often dpms goes from off to on. The cause of this problem is that when we put pm idle, I think that there is a timing problem if I do not sync. Best regards, Hoegeun > >> Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> >> --- >> drivers/gpu/drm/vc4/vc4_dsi.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/drivers/gpu/drm/vc4/vc4_dsi.c b/drivers/gpu/drm/vc4/vc4_dsi.c >> index 11702e1d9011..b0264c7b8bed 100644 >> --- a/drivers/gpu/drm/vc4/vc4_dsi.c >> +++ b/drivers/gpu/drm/vc4/vc4_dsi.c >> @@ -822,7 +822,7 @@ static void vc4_dsi_encoder_disable(struct drm_encoder *encoder) >> clk_disable_unprepare(dsi->escape_clock); >> clk_disable_unprepare(dsi->pixel_clock); >> >> - pm_runtime_put(dev); >> + pm_runtime_put_sync(dev); >> } >> >> /* Extends the mode's blank intervals to handle BCM2835's integer-only >> -- >> 2.17.1
Hoegeun Kwon <hoegeun.kwon@samsung.com> writes: > On 4/2/19 2:48 AM, Eric Anholt wrote: >> Hoegeun Kwon <hoegeun.kwon@samsung.com> writes: >> >>> There is a problem when often dpms goes from off to on. pm idle is not >>> in sync and the problem occurs. Modify pm_runtime_put from >>> asynchronous to synchronous. >> Why would we need the power domain to go to off before we try to come >> back? Any idea? Also, please specify what "the problem" is. > > Hi Eric, > > > First thank you for your review. > > There is a problem failed to runtime PM enable on DSI when often dpms What do you mean by "failed to runtime PM enable"? The pm_runtime_enable() returned an error? Have you investigated the source that error, if so?
On 4/10/19 1:24 AM, Eric Anholt wrote: > Hoegeun Kwon <hoegeun.kwon@samsung.com> writes: > >> On 4/2/19 2:48 AM, Eric Anholt wrote: >>> Hoegeun Kwon <hoegeun.kwon@samsung.com> writes: >>> >>>> There is a problem when often dpms goes from off to on. pm idle is not >>>> in sync and the problem occurs. Modify pm_runtime_put from >>>> asynchronous to synchronous. >>> Why would we need the power domain to go to off before we try to come >>> back? Any idea? Also, please specify what "the problem" is. >> Hi Eric, >> >> >> First thank you for your review. >> >> There is a problem failed to runtime PM enable on DSI when often dpms > What do you mean by "failed to runtime PM enable"? The > pm_runtime_enable() returned an error? Have you investigated the source > that error, if so? I'm sorry for the late reply. The pm_runtime_enable() is not returned because return value is void. The problem is that if the error log is not output and the DPMS off on is repeated, the screen will stop. As a result of debugging the problem, I think that synchronization is a problem caused by dsi pm_suspend and resume. So when we entered the pm runtime idle state, if used with sync, the problem does not occur. Best regards, Hoegeun <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> </head> <body text="#000000" bgcolor="#FFFFFF"> <div class="moz-cite-prefix">On 4/10/19 1:24 AM, Eric Anholt wrote:<br> </div> <blockquote type="cite" cite="mid:87ftqrno2v.fsf@anholt.net"> <pre class="moz-quote-pre" wrap="">Hoegeun Kwon <a class="moz-txt-link-rfc2396E" href="mailto:hoegeun.kwon@samsung.com"><hoegeun.kwon@samsung.com></a> writes: </pre> <blockquote type="cite"> <pre class="moz-quote-pre" wrap="">On 4/2/19 2:48 AM, Eric Anholt wrote: </pre> <blockquote type="cite"> <pre class="moz-quote-pre" wrap="">Hoegeun Kwon <a class="moz-txt-link-rfc2396E" href="mailto:hoegeun.kwon@samsung.com"><hoegeun.kwon@samsung.com></a> writes: </pre> <blockquote type="cite"> <pre class="moz-quote-pre" wrap="">There is a problem when often dpms goes from off to on. pm idle is not in sync and the problem occurs. Modify pm_runtime_put from asynchronous to synchronous. </pre> </blockquote> <pre class="moz-quote-pre" wrap="">Why would we need the power domain to go to off before we try to come back? Any idea? Also, please specify what "the problem" is. </pre> </blockquote> <pre class="moz-quote-pre" wrap=""> Hi Eric, First thank you for your review. There is a problem failed to runtime PM enable on DSI when often dpms </pre> </blockquote> <pre class="moz-quote-pre" wrap=""> What do you mean by "failed to runtime PM enable"? The pm_runtime_enable() returned an error? Have you investigated the source that error, if so? </pre> </blockquote> <pre>I'm sorry for the late reply. The pm_runtime_enable() is not returned because return value is void. The problem is that if the error log is not output and the DPMS off on is repeated, the screen will stop. As a result of debugging the problem, I think that synchronization is a problem caused by dsi pm_suspend and resume. So when we entered the pm runtime idle state, if used with sync, the problem does not occur. Best regards, Hoegeun </pre> <table id=bannersignimg data-cui-lock="true" namo_lock><tr><td><p> </p> </td></tr></table><table id=confidentialsignimg data-cui-lock="true" namo_lock><tr><td><p><img style="border: 0px solid currentColor; border-image: none; width: 520px; height: 144px; display: inline-block;" unselectable="on" data-cui-image="true" src="cid:cafe_image_0@s-core.co.kr"> </p> </td></tr></table></body> </html> <img src='http://ext.samsung.net/mail/ext/v1/external/status/update?userid=hoegeun.kwon&do=bWFpbElEPTIwMTkwNTIyMTAyNDQzZXBjYXMxcDMyMTJlODU3ZTA3NTljZDA2ZmU1MDRkMjU0MmVjNmU5YSZyZWNpcGllbnRBZGRyZXNzPWRyaS1kZXZlbEBsaXN0cy5mcmVlZGVza3RvcC5vcmc_' border=0 width=0 height=0 style='display:none'>
On Wed, May 22, 2019 at 12:24 PM Hoegeun Kwon <hoegeun.kwon@samsung.com> wrote: > On 4/10/19 1:24 AM, Eric Anholt wrote: > > Hoegeun Kwon <hoegeun.kwon@samsung.com> <hoegeun.kwon@samsung.com> writes: > > > On 4/2/19 2:48 AM, Eric Anholt wrote: > > Hoegeun Kwon <hoegeun.kwon@samsung.com> <hoegeun.kwon@samsung.com> writes: > > > There is a problem when often dpms goes from off to on. pm idle is not > in sync and the problem occurs. Modify pm_runtime_put from > asynchronous to synchronous. > > Why would we need the power domain to go to off before we try to come > back? Any idea? Also, please specify what "the problem" is. > > Hi Eric, > > > First thank you for your review. > > There is a problem failed to runtime PM enable on DSI when often dpms > > What do you mean by "failed to runtime PM enable"? The > pm_runtime_enable() returned an error? Have you investigated the source > that error, if so? > > I'm sorry for the late reply. > > The pm_runtime_enable() is not returned because return value is void. > > The problem is that if the error log is not output > and the DPMS off on is repeated, the screen will stop. > > As a result of debugging the problem, I think that synchronization > is a problem caused by dsi pm_suspend and resume. > > So when we entered the pm runtime idle state, if used with sync, > the problem does not occur. > > Sounds like not enough hystersis between disable/enable, i.e. some msleep missing somewhere. That's at least my guess, we've had a bunch of these with delayed runtime suspend of various things in i915. -Daniel > > Best regards, > Hoegeun > > > > > > >
diff --git a/drivers/gpu/drm/vc4/vc4_dsi.c b/drivers/gpu/drm/vc4/vc4_dsi.c index 11702e1d9011..b0264c7b8bed 100644 --- a/drivers/gpu/drm/vc4/vc4_dsi.c +++ b/drivers/gpu/drm/vc4/vc4_dsi.c @@ -822,7 +822,7 @@ static void vc4_dsi_encoder_disable(struct drm_encoder *encoder) clk_disable_unprepare(dsi->escape_clock); clk_disable_unprepare(dsi->pixel_clock); - pm_runtime_put(dev); + pm_runtime_put_sync(dev); } /* Extends the mode's blank intervals to handle BCM2835's integer-only
There is a problem when often dpms goes from off to on. pm idle is not in sync and the problem occurs. Modify pm_runtime_put from asynchronous to synchronous. Signed-off-by: Hoegeun Kwon <hoegeun.kwon@samsung.com> --- drivers/gpu/drm/vc4/vc4_dsi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)